Even Newer 4 to 24-cell Battery Management System (BMS)

[Nate]

Ok, I'll wait on the gate driver. I have one or two of every part arriving tomorrow so I could replace whatever needs it. I hope we can figure the part(s) out.

All the readings on the shunt resistors are the same as the battery connections at about 3.75v. Which makes sense x 16 = 60v... The reading of 16v came from the ground just below the any shunts line... I managed to blow a trace on the bottom from shorting #4 battery to shunt 3. :x Dammit!
I repaired the trace and the readings seem the same. Could that damage anything else?

The BMS is pulling .6-.8a from the charger. Were you referring to this or do I need to test the amps between each channel?
*Edit: Just tested - Each channel reads about .6a... and the LED turns off for that channel during the test.

I know how frustrating it is to troubleshot without the hardware in front of you, I work in IT. If we can't figure this out soon I may take you up on that offer. It is just so darn close... Thanks again.

 

[fechter]

Yes, it is very difficult to troubleshoot remotely.

3.75v seems too high, but odd that they are all the same. The current in all the cell circuits will be essentially the same since they are all in series. You can measure at the charger connection.

Hopefully the short circuit didn't blow anything else. I think that particular short has a low probability of toasting anything other than the traces you fixed.

The voltage across the shunt resistor should be slightly less than the cell voltage. There should be some drop in the shunt transistor. On my board, I typically see about 3v across the resistor and 3.7v across the cell. The remaining 0.7v or so should be across the emitter and collector of the shunt transistor. You can measure this by probing the inboard side of one shunt resistor to the outboard side of the one above it.

Just for fun, you could try lifting one end of any shunt resistor and see what happens. This should drop the max. current down to about 10ma. Only the cell with the lifted shunt should light up and the main LED should go mostly green.

 

[Nate]

Hopefully this will shed some light on the problem: See attached picture for details...
The voltage across the shunt resistor (A-B) is 3.5v. From inboard shunt resistor to outboard shunt resistor (B-C) is .25v. When the board is latched I get 3.5v on the transistor from emitter to collector (1-2) and about 3v between collector and base (2-3). View attachment LiFePo4-BMS-V2.3-Labled-s.jpgThese are supposed to be 512-KSA931YTA PNP Epitaxial Silicon Transistors and 512-FAN431LZXA 2.5V Linear Shunt Regulators right?

 

[fechter]

The transistors are the same ones I'm using, as well as the 431's.

It appears the shunt is fully on, but at 3.5v, it should be more than enough to fire the opto.

What is the voltage across the LED? On mine, the LED lights up fully at about 2v. It's almost like the LEDs aren't passing enough current to drive the opto. Your LEDs look a bit different than mine also, but as long as the voltage drop on them is the same, it should work. What part # did you use for the LEDs? The cell LEDs are part of the circuit and need to have a forward voltage of right around 2v for the circuit to work properly.

Also, if there's any way you can measure the voltage across R105, that would be helpful. R105 is the one closest to the LED.

 

[Nate]

The LEDs are 859-LTL1CHKFKNN. The forward voltage is rated at 2.05v.

I will get back to you on my LED and R105 voltages.

 

[fechter]

The LEDs are 859-LTL1CHKFKNN. The forward voltage is rated at 2.05v.

I will get back to you on my LED and R105 voltages.

Those seem like they should work OK. The voltage across R105 will tell us the current going into the opto.

 

[Nate]

Here are the voltages across R105:

• 1 = 0.980
  [*] 2 = 1.316
  3 = 0.933
  4 = 0.948
  5 = 0.885
  [*] 6 = 1.330
  [*] 7 = 1.354
  [*] 8 = 1.496
  9 = 0.834
  10 = 0.964
  [*]11 = 1.446
  12 = 0.978
  13 = 0.963
  [*]14 = 1.643
  15 = 0.946
  [*]16 = 1.484

 

[fechter]

Hmm.... those look about right. The opto should be firing at around 1v.

I guess another test would be to measure the voltage across R106 on a few of the highest ones.
I should have told you to measure R106 instead of R105 anyway, as this will indicate the acutal opto emitter current (brain fart). Sorry.

 

[Nate]

Here are the voltages across R106: These are in mV..

• 1 = 51
  2 = 301
  3 = 32
  4 = 38
  5 = 19
  6 = 317
  7 = 340
  8 = 470
  9 = 7.5
  10 = 45
  11 = 433
  12 = 60
  13 = 52
  14 = 608
  15 = 36
  16 = 455

 

[fechter]

OK, then it looks pretty sure that the problem is on the output side of the optos. Something must be holding the any shunts line high, and there aren't too many places left that could do that.

There's not much left to check at this point. Maybe it's time to try replacing the gate driver.

 

[Nate]

No joy... Replaced the U2 gate driver, no change. I has to be the any shunts line... What did we miss?

 

[fechter]

I'm running out of ideas

On my board, the R106 voltages are very consistent, running around 520mv when throttling. The any shunts line is around 1.9v. From this, I assume that it won't start throttling until the R106 voltage gets up to 520mv. It seems like you're not quite getting enough charger voltage to trigger the optos, but you tried boosting it before and that didn't seem to do it.

You could try placing a jumper across two cell circuits (above cell 5) breifly and see if the main LED does anything. By jumping two cells, the voltage on the remaining cells will be about 7v higher. Don't run it with the jumper for more than a few seconds if the LED stays red.

With the power off, measure the resistance of R101 and R102 on a few cells to verify the divider resistors. If the voltage set point (determined by the divider) is too high, then the circuit will take more voltage to start throttling. Edit: measure with the probes in both directions as the readings seem to be affected by the LM431 being attached.

 

[Malcolm]

Just a thought: is your voltmeter definitely accurate Nate? A while back I started getting some weird voltage readings, and it took a while before I realised the battery in my meter was dying and giving gradually climbing readings. Could it be that your charger isn't supplying as high a voltage as you think?

 

[Nate]

That is definitely a good idea and worth checking out. I tested in both meters with a separate battery checker and one, the cheap Ideal, had barely low batteries. I changed them just in case. The batteries in the Mastech are good. The readings in some places are up and down a bit. Maybe because of the capacitors?

I can't tell you guys how much I hope each idea fixes this problem. I'm starting to think the problem is more compound. The high and low voltages on some channels makes me think the problem is around there. Notice channel 14? It is very high in both tests. Channel 9 and 5 are very low...

I have tested all the resistors several times but I will check R101 and R102 again and together.

That extra 5v on the any shunts line is bothering me too. I need to figure out were that is coming from. I will work on it more tonight.

Thanks for the help guys. Keep the suggestions coming...

 

[fechter]

Thanks Malcolm, good thought.

It does seem to behave like it's just not getting enough voltage to trip the optos.

I would still suggest trying the jumper across two or three cells for a few seconds to see what happens. I don't think it can fry anything for a short time.

 

[Nate]

I did the jumper and it just turned off the LED for that channel. It would not lock.

The R101 and R102 seem very close per channel. I'm not sure how much of a difference a few k ohms would make...Not that familiar. For example the 180k is around 175k. I can get exact reading if you think this is still the problem.

I decided to charge up my battery and just unplug it once one channel got up to 3.68v. So I left R6 "lifted" and let several channels get up to voltage about 3.7v and it turned itself off...and then back on. It was throttling, somewhat... So I hooked up two 20k and the 4.7k and it really started throttling. Off and on, off and on. It started to make my charger fan slow down. Is that normal? I feel like it is not good for the charger. Anyway I stopped the throttling at that point and hooked up the R6 4.7k until all but 1,3,5,7, and 9 were up to 3.7V. (I would let it charge until one channel was up to 3.75 then disconnect to drain the high channels back to 3.68v. It didn't take very long.) Now I am single charging those last channels using a HYPERION EOS 0610I DUO charger to get the pack balanced. I really think it is that extra voltage on the any shunts line that is the problem. Please let me know what you think... I will let you know if I can get it to lock with all the channels balanced.

 

[fechter]

When you put the jumper across a few cells, it won't lock green, but you should see it become less red or start blinking red/green as long as the jumper is on. Did you see any change in the LED color?

The on/off behavior is correct.
Normally it turns on and off so fast you can't see it and the main LED looks yellow or somewhere between red and green. If it blinks slowly, that is OK from a charging standpoint as long as the cells maintain proper voltage. My cheap chinese SLA charger does something similar toward the end of balancing. The speed of oscillation is largely determined by your cell impedance. Larger cells will run slower.

 

[Nate]

Yes I see that now. The main LED gets lighter red the more cells I jump.

Here is the hard data, all numbers in kohms:
Channel - divider, divider other way, R102, R102 other, R101, R101 other

• 01 - 250.5 , 96.5 , 177.1 , 122.4 , 74.6 , 70.9
  02 - 250.5 , 96.1 , 177.5 , 122.9 , 74.6 , 70.8
  03 - 250.7 , 94.7 , 177.9 , 122.1 , 74.5 , 70.7
  04 - 250.6 , 98.7 , 177.3 , 124.5 , 74.6 , 71.1
  05 - 250.5 , 94.5 , 177.7 , 121.8 , 74.6 , 70.6
  06 - 250.5 , 93.8 , 177.5 , 121.4 , 74.8 , 70.7
  07 - 250.7 , 96.4 , 177.7 , 122.9 , 74.4 , 70.7
  08 - 250.4 , 94.6 , 177.7 , 121.6 , 74.6 , 70.7
  09 - 250.2 , 94.9 , 177.5 , 121.7 , 74.5 , 70.7
  10 - 251.1 , 94.7 , 178.2 , 121.8 , 74.6 , 70.7
  11 - 250.7 , 97.8 , 178.0 , 123.7 , 74.8 , 71.2
  12 - 250.5 , 95.1 , 177.7 , 121.9 , 74.6 , 70.7
  13 - 250.6 , 95.2 , 178.0 , 122.0 , 74.4 , 70.6
  14 - 250.2 , 96.8 , 177.5 , 122.8 , 74.7 , 71.0
  15 - 250.4 , 95.1 , 177.6 , 121.9 , 74.6 , 70.7
  16 - 250.8 , 97.8 , 177.9 , 123.7 , 74.6 , 71.0

The tests above are worth 96,000 opinions!

 

[steveo]

Hey Everyone,

I just wanted to ask what voltage i should set my psu on to charge 24s of a123? I have my psu set at 87.6 or 3.65v per cell..

I never see the bms switch to the green light completely .. all cells when i checked them are around 3.67-3.65 .. do i just need to let it balance longer ? increase or decrease voltage .. I've seen the light go amber today .. but then flip back to red ..

-steveo

 

[Malcolm]

steveo
My charger was originally set for 43.8V for 12 cells (3.65V, same as your PSU), but I found that was just a touch too low. I've tweaked it to 44.3V and that made all the difference, so I'd suggest trying 88.5V. I plan to try cutting back the voltage again when my pack has done a few cycles, as I suspect it won't need the extra once the cells have got to know each other...

 

[Nate]

When you put the jumper across a few cells, it won't lock green, but you should see it become less red or start blinking red/green as long as the jumper is on. Did you see any change in the LED color?

So do I have a couple problems going on here? Could they be linked? Could it be that the extra voltage on the any shunts is linked to not getting enough voltage to trip the optos. A leak somewhere? Should I start pulling up Diodes? I want to keep trying to fix it, just don't know what to try next... Please let me know if I can get you any more readings.

 

[fechter]

So do I have a couple problems going on here? Could they be linked? Could it be that the extra voltage on the any shunts is linked to not getting enough voltage to trip the optos. A leak somewhere? Should I start pulling up Diodes? I want to keep trying to fix it, just don't know what to try next... Please let me know if I can get you any more readings.

Well, the divider resistors look good.
If the main LED is changing color when you jump cells, that indicates the throttling circuit is working. Everything seems to be working...

Next you might try going back to using all channels and try to increase the charger voltage a bit (enough to see a slight color change in the main LED). Measure the voltage between cell taps and maybe across the power resistors. I found I got goofy readings with no cells attached, but if I put a capacitor across the meter leads (100uf - 1000uf) then the readings were more in line with expected. If you don't have a capacitor, just measure but keep in mind the readings will normally be high.

 

[AndyH]

... So do I have a couple problems going on here? ...

Nate - have your checked your TC54s? They are directly across the cell and if damaged can keep a channel from shunting... A long shot, but... I damaged one so badly that it not only kept the cell from charging, it continued to drain the cell until the BMS was disconnected.

 

[Nate]

Everything seems to be working...

So Close...

Next you might try going back to using all channels and try to increase the charger voltage a bit (enough to see a slight color change in the main LED). Measure the voltage between cell taps and maybe across the power resistors.

I increased charger voltage, using two 1.5v batteries, to 63v and the light stays solid red.
These tests are with 60v on the charger:
Channel - Taps - Resistor - - Channel - Taps - Resistor

• 1 - 3.73 - 3.50 - - 9 - 3.75 - 3.51
  2 - 3.72 - 3.48 - - 10 - 3.74 - 3.50
  3 - 3.72 - 3.48 - - 11 - 3.73 - 3.49
  4 - 3.71 - 3.48 - - 12 - 3.75 - 3.51
  5 - 3.77 - 3.53 - - 13 - 3.74 - 3.50
  6 - 3.75 - 3.51 - - 14 - 3.75 - 3.50
  7 - 3.73 - 3.49 - - 15 - 3.73 - 3.49
  8 - 3.73 - 3.48 - - 16 - 3.73 - 3.49

have your checked your TC54s? ...damaged one so badly that it not only kept the cell from charging, it continued to drain the cell until the BMS was disconnected.

Hi Andy. I'm using the TC54VN2102EZB-ND. I haven't checked these yet, other than LVC. What is the best method? I can tell you that all channels are charging. Without the battery connected I get all the LEDs to light and each channel reads as stated above. With the battery all LEDs come on at about 3.68v and if I leave it on for awhile they get up to about 3.73v. I haven't noticed any draining yet. I will look for that.

For now I'm just watching while charging my battery and I unplug it when all the LEDs are on. They are all on within a minute of the first LED.
***Edit***: My second charger just died trying to charge this battery. The Soneil 4808SRF was the first to go. It was bought at Batterymart and has some obvious rework done inside. The other is PLC4806B China Export. I got it from a local dealer. I expect he will cover this one. What is a good charger for this BMS that will last longer than a month????? Thanks everyone.

 

[GGoodrum]

I just updated the website with the latest version, which is now v2.5. Nothing magor, just a preventative change Richard made to make the FET gate control a bit more stable, and a switch to non-polarized MLCC caps. Also changed, slightly, is the layout. We spread things out a bit, and fixed some lableing. It should be a bit harder now to have solder bridges in the tight areas. That is about the extent of the changes. I've updated the instructions, and the BOM files.

-- Gary